Usage
Task guides for each interface. They share one model:
- Number a sequence: give every residue a position in the scheme you chose (
number). - Segment it: split it into FR1–FR4 and CDR1–CDR3, plus the residues before and after the
domain as
prefixandpostfix, so the pieces always join back into the input (segment). - Do either for every domain in a sequence, such as both domains of an scFv
(
number_domains,segment_domains, or--all-domainson the command line).
You set the chains to consider, the scheme and the minimum confidence once, when you build an
annotator; Concepts explains each choice. A sequence that can't be numbered gives
a result with error set rather than stopping the batch; see Errors.
| Guide | For |
|---|---|
| Command line | FASTA or plain-text files, shell pipelines; TSV and JSON output |
| Python | immunum.Annotator, one sequence at a time |
| Polars | immunum.polars, large datasets inside Polars queries |
| JavaScript | The npm package for Node.js |
| Rust | The immunum crate |
| Errors | What every interface raises or returns when something goes wrong |
| Upgrading to 2.0 | What can break when coming from 1.x |
Installing: Getting started. Every parameter and field: the API reference.